A Ramble On Microphones, the DJI Mic 2, Paul Garrity, Ruined Videos, Etc...

[@ggdfggdfgdffgfddg34]

raw video, is it worth it? or is compressed video in 264 the same? And most importantly, why is the dynamic range always compressed in raw video?

[@MaxoticsTV]

HAHA, I can never pronounce your name right ;)

[@ggdfggdfgdffgfddg34]

@@MaxoticsTV my name is Eugene) what about raw vs standart ptofil?

[@MaxoticsTV]

@@ggdfggdfgdffgfddg34I thought you were winding me up? If you're not, can you re-phrase the question? Thanks! In the end, we get 8-bit dynamic range, whatever we record in.

[@ggdfggdfgdffgfddg34]

@@MaxoticsTV I understood about 8 bits, but when recording in raw according to cined tests, in particular Nikon raw, we can notice a decrease in the dynamic range. I do not understand why this happens, raw gives us a noisier image, but the dynamic range in theory should not decrease, but the program records clean steps. And if you believe the tests for exposure latitude, then gh7 gives 9, stops of exposure latitude. Even more than in Nikon z9, the final question is if raw gives more noise, why shoot in it, if on average it gives 2 steps less DD?

[@MaxoticsTV]

I don't pay attention to the "dynamic range" claims because they are incomplete. In 12-bit RAW, you should be able to discern a range of brightness between 1 and 4,096 (highest values in 12-bit). Photographers (filmmakers) think in terms of stops. At which stop can I see detail. But it's generalized, not scientific. It doesn't take into account color, or noise. Why? Because there is no agreed-on standard between the manufacturers. Long ago, they had to accept ISO or ASA because photographers had to have a benchmark when buying film and paper and processing etc. But those definitions were based on black and white photography. What I'd need to see to listen to "dynamic range" claims is an agreed on test for noise and then each stop would have an addition value of noise. So it might be 10 stops with a range of 0 to 30% color noise, or 12 stops with 25% noise, etc. Using a gray scale to derive "dynamic range" is pointless to me. At the end of the day, it doesn't matter how many stops you have if the ones at the far end are too noisy to use.

[@davidfirth1]

Just can't walk away, though I've tried. You seem like a nice guy. You're not that much older than me. So, maybe some affinity bias on my part. Not sure why the algorithm suggested your channel to me, but I've watched a few and you are really shaking your fist at the sky on things that, as an electronics engineer, I can say you do not seem equipped to accurately analyze. You also seem very staunchly stuck in your opinion. A better approach would be to find someone who works in digital signal processing hardware, ask how this is done, what the assumptions are, what gotchas there might be, and keep the measure an electron, the river is never the same, and side rants on vaccination out of it (I did spend 10 minutes looking up some of your back and forth on the ASR forums, too - oh boy). The practical result is that 32 bit float is acting as a really good safety track when everything else in the chain is working as designed. There's enough headroom in the 32 bit float processed end product that it takes 20dB-30dB of 'gain' addition to bring up the result in an audio editor to a workable level. More than enough to quantify the loudest and softest sounds that one would want to capture. There is no risk of floating point 'imprecision' being a problem with this use case; we're not running weather models or trying to insert a satellite into Jupiter's planetary orbit over an exact point. Those who are using 32 bit float can, if they don't understand or don't have the extra time to 'do it right' per the old ways, relax parts of their audio workflow. Maybe that's using technology as a crutch. Personally, I don't use 32 bit float often, because I'm not a fan of the extra workflow and I sound check and adjust gain in my AV side gig -- and if I interact with contracted folks, most professionals request a traditional 24 bit workflow. But, I get why 32 bit float is working for people who choose to use it. And marketing people will always find their flavor of spin. This isn't the techie's fault and doesn't invalidate what the engineering solution does. So, ignore or contemplate. Your call. I can always just tell the algorithm not to show me any more.

[@MaxoticsTV]

Thanks for the comment. I disagree that I am "very staunchly stuck" in my opinion. I feel more like a sane person trapped in an insane asylum where nothing I say will get a sane response--in my view. But hey, I'm insane, right? I am not an electronics engineer. I have never claimed otherwise. But that doesn't mean, again, that I'm going to agree with an "electronic engineer" who claims that 32-bit float reduces clipping without proof. Most of all, if I AM WRONG, and someone explains how, I will be the first to admit it. I am not looking to be wrong. What you're basically saying here is I say a lot of crazy stuff, tilting at windmills, therefore I must be wrong because those who "shake their first" are always wrong. How can I reply to that? So let's put that aside. I've written all the manufacturers. I've asked others, as you saw, on the ASR forum. No one has given me as scientific or technical reply. They talk about "20-30db" of gain this or gain that. Those are just numbers. What are they explaining? What physical effects do those numbers represent? But fine, one doesn't think I'm qualified to understand the engineering. Then show me a controlled scientific test where you take a recorder in 24-bit and one in 32-bit float, where there is no limiter technology between the pre-amp and ADC, no DSP tech applied to the data stream, and demonstrate, PROVE, that 32-bit float reduces clipping? If it's so good, why hasn't there been proof by the manufacturers? Or why can't you do it? Why don't YOU use 32-bit float? If I'm wrong, you'd think you'd take your own advice? ;) Again, I'm all ears! I'd LOVE for you to put me out of my insanity here. You can tell the algorithm "not to show me anymore". Oookkaaay ;) Who's talking crazy now? Or you can show me the science that proves that 32-bit float records better fidelity than 24-bit. Here is my essay. BEBUNK away! But a word of warning. I AM WRONG a lot. It's painful. Can you take it, if you're wrong? maxrottersman.medium.com/the-three-types-of-microphone-clipping-explained-d23ad13a60f7?source=friends_link&sk=2741a5e36687099eb740aa92b80fa8c1

[@MaxoticsTV]

Knowing that you "never want to give up on learning and exploring" I made a video for you. You can point out my misunderstandings since you're an "electronics engineer" and I am not. kzbin.info/www/bejne/qpW3cneofMqme5Ysi=CCAGXg7D4Z1foJb5 I look forward to learning from an expert!